Conformal Prediction using Ridge2 • ahead

Ridge2 is a uni/multivariate nonlinear probabilistic time series model originally presented in Multiple Time Series Forecasting Using Quasi-Randomized Functional Link Neural Networks.

library(ahead)

## Registered S3 method overwritten by 'quantmod':
##   method            from
##   as.zoo.data.frame zoo

Univariate

AirPassengers

plot(ahead::ridge2f(AirPassengers, h=30L, lags=20L, type_pi="conformal-split"))

plot(ahead::ridge2f(AirPassengers, h=30L, lags=20L, type_pi="conformal-block-bootstrap"))

plot(ahead::ridge2f(AirPassengers, h=30L, lags=20L, type_pi="conformal-bootstrap"))

USAccDeaths

plot(ahead::ridge2f(USAccDeaths, h=20L, lags=10L, type_pi="conformal-split"))

plot(ahead::ridge2f(USAccDeaths, h=20L, lags=10L, type_pi="conformal-block-bootstrap"))

plot(ahead::ridge2f(USAccDeaths, h=20L, lags=10L, type_pi="conformal-bootstrap"))

Multivariate

obj <- ahead::ridge2f(fpp::insurance, 
                      h=10L, lags=2L, 
                      type_pi = "conformal-split")


plot(obj, "Quotes")

plot(obj, "TV.advert")

obj <- ahead::ridge2f(fpp::insurance, 
                      n_hidden_features = 0L, 
                      h=10L, lags=1L, 
                      type_pi = "conformal-block-bootstrap")


plot(obj, "Quotes")

plot(obj, "TV.advert")

obj <- ahead::ridge2f(fpp::insurance, 
                      h=10L, lags=2L, 
                      type_pi = "conformal-bootstrap")


plot(obj, "Quotes")

plot(obj, "TV.advert")

plot(obj, "Quotes", type = "dist")

plot(obj, "TV.advert", type = "sims")